The paper introduces a novel computer vision task called Game State Reconstruction (GSR), which aims to track and identify athletes on a football pitch using video inputs from a single camera. The authors formalize GSR as a task that includes pitch localization, camera calibration, athlete detection, re-identification, tracking, role classification, team affiliation, and jersey number recognition. They release the SoccerNet-GSR dataset, consisting of 200 30-second video sequences with over 9.37 million line points for pitch localization and camera calibration, and over 2.36 million athlete positions with their respective roles, teams, and jersey numbers. To evaluate GSR methods, they introduce the GS-HOTA metric, which accounts for the specificities of GSR tasks, such as the additional target attributes and the 2D points provided instead of bounding boxes. The paper also proposes the GSR-Baseline, an end-to-end pipeline for GSR, and evaluates its performance using the GS-HOTA metric. The experiments highlight the complexity of GSR and the interdependencies among its subtasks, suggesting areas for future research.The paper introduces a novel computer vision task called Game State Reconstruction (GSR), which aims to track and identify athletes on a football pitch using video inputs from a single camera. The authors formalize GSR as a task that includes pitch localization, camera calibration, athlete detection, re-identification, tracking, role classification, team affiliation, and jersey number recognition. They release the SoccerNet-GSR dataset, consisting of 200 30-second video sequences with over 9.37 million line points for pitch localization and camera calibration, and over 2.36 million athlete positions with their respective roles, teams, and jersey numbers. To evaluate GSR methods, they introduce the GS-HOTA metric, which accounts for the specificities of GSR tasks, such as the additional target attributes and the 2D points provided instead of bounding boxes. The paper also proposes the GSR-Baseline, an end-to-end pipeline for GSR, and evaluates its performance using the GS-HOTA metric. The experiments highlight the complexity of GSR and the interdependencies among its subtasks, suggesting areas for future research.